The Atlantic Meridional Overturning Circulation (AMOC) is a key component of the climate system, regulating heat and freshwater transport across the globe. Its stability is of particular concern because an AMOC weakening or collapse could trigger abrupt and potentially irreversible climate shifts. A critical region for this stability is the North Atlantic Subpolar Gyre (SPG), where deep convection hepls drive the AMOC. Yet, the literature presents contrasting explanations for multistability of convection in the SPG: the continuous horizontal box model of Bastiaansen suggests physically meaningful multistability, whereas the one-dimensional column model of Den Toom points to spurious multistability.

This thesis investigates whether the multistability observed in complex ocean models reflects genuine physical processes or arises as an artifact of modeling techniques. To bridge the contrasting perspectives by Bastiaansen and Den Toom, an overarching two-dimensional model is developed that contains elements of both approaches, allowing reproduction of each limiting case and systematic exploration of their combined effects.

Analysis shows that multistability in Bastiaansen’s model depends critically on parameter choices, disappearing when physically realistic parameters are used. In Den Toom’s model, multistability vanishes with increased vertical resolution, confirming that it is a numerical artifact. Since climate models lack sufficient vertical resolution, this artifact is relevant in practice.
The overarching model demonstrates that a combination of both mechanics can introduce additional multistability on top of multistability similar to Den Toom, but these features vanish with higher horizontal resolution.

Overall, the results indicate that the multistability observed in these idealized ocean models is spurious, arising from unrealistic parameters or insufficient numerical resolution rather than reflecting true ocean dynamics. These findings highlight the importance of physically justified parameterizations and sufficient model resolution in interpreting bifurcation structures and potential tipping points in climate models.